So you’ve been reading, absorbing, maybe even side-eyeing that AI wave from the shoreline. You’ve seen ChatGPT do its thing. You’ve watched people automate full workflows, launch apps in a weekend, or make AI-generated rap battles between Aristotle and Batman. You’re intrigued. But deep down, you might be thinking:
- “I’m not technical enough.”
- “I don’t know where to start.”
- “What is LangChain?”
Let’s fix that by helping you leverage the API in an LLM I spoke about earlier to build your own AI-powered application, with just prompts! No coding experience required.
This chapter is your permission slip, your map, your flashlight — and most importantly, your invitation to build. Welcome to vibe coding.
Vibe coding is building with AI when you don’t know everything, and don’t need to.
It’s learning by doing. Assembling blocks. Letting the AI help you write the parts you don’t know. You're not writing code in the dark — you're collaborating with it. Think of yourself as a director working with a super-competent assistant who just needs instructions.
You don’t start by reading a 600-page machine learning book.
You start by making something you care about.
WOW! SO DO WE STILL NEED DEVELOPERS?: Yes
LONGER ANSWER: A lot is discussed about AI replacing all developers, so I want to be clear on my position here. Yes, AI can be a great resource to write code and even write basic applications like the one we are writing here. However, experienced developers are still required to leverage AI for building complex systems and know how systems and architectures operate. Like I said earlier, you can’t just tell AI to write Salesforce and treat AI like an incapable tool because it doesn’t spit out a complicated app that works out of the box. Yes, AI is mighty and grows more capable by the day, but it’s a capability best leveraged by experienced developers to augment their capability because software, systems, and technology are all very complex and that background is vital to leveraging AI to build effectively in any complex environment.
Using AI to build a simple app, like we are going to demonstrate here, is a great way to dive into what more and more technology experts are embracing for their day-to-day work, and AI does get more capable by the day, so this will change as the technology improves. Right now, though, and likely for some time into the future, technology domain experts across the entire stack will still be needed to help instruct AI on specifically what to do and how to do it. More importantly, AI tools just don’t build everything in a single prompt – it’s a process to build iteratively using vibe coding, and for more complex use cases, it still takes a lot of time to get things built correctly. There’s no denying, though, that as AI becomes more capable, it’ll be able to do more with less, so come back here in a year and I have to update this.
So, now back to vibe coding your first app!
Before you dive into app-building, you’ll want an environment that actually helps you code with AI. Cursor is exactly that, and is a great tool to start using for Vibe Coding.
Cursor is a modified version of Visual Studio Code (an IDE or Software Development tool that developers of all kinds use) that’s supercharged with AI built right in, so you can use AI to build alongside your prompts. It’s great for vibe coders like you - you write partial thoughts, and the AI helps you turn them into real apps, from testing to running.
How to Set It Up
- Go to https://www.cursor.so and download the version for your OS (Mac/Windows/Linux).
- Install and launch it, picking a folder on your computer where you want to store the project folder.
- Create a free Cursor account (you can use GitHub/Google).
- Connect your OpenAI API key:
§ Sign in to https://platform.openai.com/account/api-keys
§ Fill out the form to add your credit card and put $5 in the account, which allows you to use “tokens” (using a single token lets you pay for the LLM to compute a single transaction, with $5 getting you pretty far to start).
- Generate a new key and paste it into Cursor’s settings, to leverage the AI inside of cursor.
- Generate a second key called “TestApp” which you’ll need after following the instructions below.
- Setup your dev environment (lets start with Python), and for this, you can actually ask Cursor in the AI chat interface to set it up for you if you’re not comfortable with that. It’s a conversation with Cursor, it’ll ask questions and if you don’t understand the question, tell Cursor that as it’ll break things down for you. Once it says your Python development environment is set, it’ll likely prompt to set up a virtual environment to begin coding in as well, which is fine to do.
- If you get stuck/lost, just tell Cursor that – it’s a pretty amazing instructor/teacher. I could write an exhaustive step-by-step here, but we are using AI now! So, keeping it vague enough that AI can pick up the slack based on the help you need or questions you have.
Tip: Cursor supports both GPT-4 and Claude, and will integrate with your terminal and GitHub repo too (which is where devs publish their code into the cloud to share with others). This is not a toy to be clear - it’s a production-level AI coding assistant, but it’s easy to leverage to start getting comfortable with AI-based coding and can grow as your interest and ability grow as it’s used by developers of all kinds globally today.
Don’t build something generic. Build something you’d use. Here are some quick prompts to unlock ideas:
- “I want to create an app that reads my PDFs and lets me ask questions about them.”
- “I want to automatically summarize YouTube videos I bookmark.”
- “I want a web app that turns my daily notes into a to-do list.”
- “I want to text a bot that recommends guitar tabs based on how I’m feeling.”
Let’s pick one to walk through to demonstrate what a prompt for Cursor might look like (feel free to modify as you see fit):
Project Prompt for Cursor (Modify then Paste this Into Cursor AI box):
Create a simple AI chatbot using Streamlit that allows me to paste a page of notes into a chat window, then push a button titled “Submit” to send that note information to the AI agent to process and prompt to ask the agent to ready my notes and share its observations based on what was submitted in a read-only chat window beneath the screen where my notes were pasted. The application should only have two boxes, one for the text input for pasting text, and the second as a read-only window to post the output from the AI agent built into the application. Along with a button to submit my notes for the agent to ingest and analyze, there is a second button to clear the AI session and remove all the text from both boxes. Please also include a window on the left side of the application where I can post the API key into the application, so that I don’t need to pre-populate a key into a secret or .env file, as that may be too technical for me. Also, please write and execute a test script to validate the functionality of this application before I execute the streamlit session to launch the application. I will use OpenAI for this application, and I want you to use GPT-4o for this project. Please also write the requirements.txt file as well as generate a readme file for this project.So there’s a lot there, and likely will have to be revised perhaps once Cursor goes and builds a first version of the application.
Once Cursor does its thing from the AI chat interface, you’ll see a bunch of coded files appear on your screen as Cursor writes the code based on the prompt above. Once that’s done, tell Cursor to execute the test script to ensure it wrote what you wanted it to, as it’ll catch it’s own errors that way. Once Cursor is done doing its testing and fixes, ask Cursor to launch the streamlit application on your computer. You should then see it spin up a window that is the application, and allow you to do just what you asked. If something doesn’t work or there’s an error, copy the error, paste it back into the Cursor, and tell it to fix the error. Once that’s done, relaunch until you don’t get an error.
Once you’ve gotten your app to work, the path forks into infinite roads, and you can have cursor keep working on your code once you’re happy:
- Ask Cursor to add a file upload so others can drop in documents.
- Add llama-index for smarter context parsing.
- Deploy the code to GitHub (once you set up an account and a private “repo”)
- Hook it into Twilio to text your chatbot.
- Use ElevenLabs to make it speak.
Each step you take gives you more confidence. AI becomes less magic and more of a co-pilot. This is a lot for someone just starting out, though, so if you’re lost, you guessed it! Use AI to post questions/screenshots/issues and grow as you tinker.
You're not here to memorize TensorFlow syntax (or even know what TensorFlow is)
You're here to learn the rhythm of AI creation — to vibe your way into building, so this is really to help you understand how AI works beyond the ChatGPT interface. This is how the next wave of founders, artists, educators, and farmers are going to work. Not from technical mastery alone, but from curiosity and collaboration with machines and understanding where AI can do what and how.
The world is changing fast. But you’re not behind. You’re right on time.
Buildspace – Projects to learn by building
FlowGPT – Community of prompt engineers
AI isn’t the future—it’s the now. Building with it doesn’t take a CS degree—it takes curiosity and momentum.
So go ahead: spin up Cursor. Dream something weird. Build it fast. Break it often. Let the AI help. Let yourself play. And then share it with the world.
You’re not just watching the future anymore—you’re coding it! This doesn’t mean you need to go any further if development isn’t your thing, but you understand now how this works for folks beyond those publicly facing chat websites.
Click on the Next Chapter Below To Keep Going
Introduction: A Company Strategy for AI: How companies develop a strategy to leverage AI, and change in a way that maximizes AI effectively
Chapter 1: A Company Strategy for AI: How companies develop a strategy to leverage AI, and change in a way that maximizes AI effectively
Chapter 2: Use AI (more) effectively as a Startup: How startups begin to leverage AI and avoid turning into a failed science project
Chapter 3: Becoming an AI Native: How individuals leverage AI, and become an “AI Native”
Chapter 5: Surfing the AI Tsunami: Where We Are, What We've Learned, and What Comes Next
No posts
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.